RESEARCH
435
June 2014, Vol. 104, No. 6Tick-box admission forms improve the quality
RESEARCH
436
June 2014, Vol. 104, No. 6 Human error is a significant problem in complexhuman organisations such as the aviation and nuclear power industries, and in modern trauma systems.
However, the aviation and nuclear power industries have used insights provided by modern error theory to develop enviable safety records. To date, the healthcare system has not achieved a comparable record.[1-3] The challenge to healthcare managers and clinicians is to use modern error theory to improve the quality of healthcare systems. Modern error theory provides insights into the evolution of error in healthcare systems by recognising that it is predominantly the system, rather than the individual, that fails.
While individuals may make mistakes, it is the system that allows human error to affect patient care. A robust system directs care along certain desired pathways. If care does not follow the appropriate pathway, a robust system will autoregulate to redirect care down an appropriate pathway. If a system is not robust, it is possible for an individual to override it and violate protocols. Violations remain a significant cause of human error in healthcare systems.
A study from the University of Pittsburgh demonstrated a correlation between inadequate documentation of prehospital care and mortality.[4] The authors reviewed all emergency medical service records for 2002 and 2003 in King County, Washington, USA.
Multivariate analysis demonstrated that failure to record one or more physiological parameters at the scene of the injury predicted an increased risk of death.[4] The authors concluded that inadequate record keeping reflected poor care. They rejected the hypothesis that the severity of the pathology treated by the prehospital staff meant that poor documentation merely reflected lack of time to make appropriate notes, and concluded that poor documentation is a proxy marker for poor care. We have previously audited the quality of documentation of trauma patients in our institutions and found it to be inadequate.[5,6]
Objective
We set out to address this deficit, and in light of the modern understanding of human error, sought to implement a standardised tick-box style admission form that would fulfil the dual role of improving the admission documentation of surgical patients and creating decision nodes to actively direct care down appropriate clinical algorithms (Appendix 1, available in the online version of this article).[7,8] By way of example, after making the admitting clinician document basic physiological data, a ‘yes’ or ‘no’ tick-box asking a clinical question such as ‘is shock present?’ or ‘does the patient require active rewarming?’ was included. This was intended to force a clinical decision and prompt an appropriate clinical response.
Methods
The study assessed the impact of the tick-box admission forms on patient safety by reviewing the quality of the data recorded and the impact of the forms on patient care. Before implementation of this intervention, all admission assessments were performed without any preprinted standardised rubric and it was our impression that the general quality of admission documentation was below an acceptable standard.
We audited 100 consecutive admissions before the introduction of this intervention. The issue of establishing an adequate bench mark for quality of documentation was discussed among senior departmental colleagues. We collectively came to a con sensus that the following 29 criteria should be present in the assessment of any emergency surgical admission: admitting doctor’s first and last name; patient’s name and
surname; a clear definition of the acute surgical pathology; time and date of clinical assessment; clarification of any significant previous medical history; clarification of any significant previous surgical history; clarification of any known allergies; clarification of any significant social history; pulse rate; blood pressure; respiratory rate;
saturation of oxygen in haemoglobin (PaO2); core body temperature;
findings on examination of the central nervous system; findings on examination of the cardiovascular system and lungs; findings on abdominal examination; use of adjuncts during resuscitation; the type(s) of resuscitation fluids utilised; the volume of resuscitation fluids utilised; urine output volumes following resuscitation; require- ment for antimicrobials; requirement for analgesia; laboratory investigations utilised; inter pretation of laboratory investi gation results;
imaging investigations utilised; interpretation of imaging investigation results; communication with senior surgical staff; definitive clinical assessment; and definitive management plan.
The tick-box clerking form was des igned by the authors. It was presented to all members of the surgical department using Microsoft PowerPoint with digital projection, together with hard copies of the form. The presentation involved a lecture on error theory and the importance of standardisation of accurate documentation (for the purposes of quality improvement), followed by a thorough orientation in the use of the form. This document was then implemented as departmental policy for the admission of all surgical patients. Medical doctors were the only staff permitted to admit surgical patients, with the demand that the tick-box form be completed following initial patient examination and stabilisation.
No other ancillary staff member was involved in the exercise of using the form.
All surgical patients admitted to the Dep artment of General Surgery at Grey’s Hospital, Pietermaritzburg, KwaZulu-Natal, South Africa, were included in the study. We categorised the admission process into the following discrete steps based on current approaches to the acute management of trauma and sepsis: resuscitation (airway, breathing, circulation, core temperature, and neurological deficit), response to resusc itation (urine output), drugs used (antibiotics and analgesia), and need for added special investigations. The ten decision nodes illus trated in Table 1 were incorporated into the clerking sheet.
The study assessed the impact of this tick-box admission form on both the quality of the admission documentation and the clinical behaviour of the admitting clinician.
One month after implementation of the system, 100 tick-box admission forms were audited and compared with the previous method of admission in respect of quality. The authors performed the audit. We assessed the impact of the decision nodes on clinical behaviour by classifying responses as either compliant (tick-box was selected) or not compliant (tick-box was not selected). Thereafter, we analysed whether the compliant nodes were accurately or inaccurately selected using the following classification:
• compliant + accurate (pathology present and appropriately recognised)
• compliant + inaccurate (pathology present without recognition or intervention).
Results
Use of the tick-box admission form resulted in a significant improvement in the quality of recorded data (unpaired Student’s t-test; p=0.0006). Table 2 compares the quality of the data recorded before and after the inter vention, and illustrates improved data entry for all parameters. How this affected clinical care is less Appendix A is available online at http://dx.doi.org/10.7196/SAMJ.7673
RESEARCH
437
June 2014, Vol. 104, No. 6certain; Table 3 summarises the analysis of the decision nodes.
Compliance was good for status of the airway, the need for supplemental oxygen and the haemodynamic status of the patients.
It was poor in terms of assessing adequacy of urine output, the need for antibiotics and analgesia, and the need for review of blood results. However, despite compliance with completion of decision nodes, the interpretation of basic clinical data was not consistently correct. Six patients in this cohort were shocked on presentation: in one case the data were not recorded, and in four cases the data were recorded but the doctor failed to recognise the pathological condition. Similarly, in three patients with a core body temperature <35oC, the clinician did not recognise that active rewarming was indicated. No patients who required antibiotics were administered the appropriate drug, and 14 who required analgesia were not given it, despite the decision node Table 1. Decision nodes
Process Clinical data Decision node
Resuscitation Airway Is an emergency airway required?
Resuscitation Arterial oxygen saturation (SaO2)
Is supplemental oxygen required?
Resuscitation Blood pressure and pulse
Is shock present?
Resuscitation Core body temperature
Is active rewarming of patient required?
Response to resuscitation
Urine output Is urine output normal or low?
Drugs Indication for
antimicrobials
Are antibiotics required?
Drugs Pain Is analgesia required?
Investigations Arterial blood gas Is the arterial blood gas normal or abnormal?
Investigations Full blood count Is full blood count normal or abnormal?
Investigations Urea and electrolytes Are urea and electrolytes normal or abnormal?
Table 2. Quality of the data recorded before and after the intervention
Resuscitation process Clinical data A (pre-intervention, N=100), n B (post-intervention, N=100), n
Resuscitation Respiratory rate 22 80
Resuscitation Oxygen saturation 45 84
Resuscitation Temperature 30 67
Resuscitation CNS examination 56 85
Resuscitation Type of fluid 16 84
Resuscitation Volume of fluid 9 46
End-point of resuscitation Urine output 5 19
Drugs Antibiotics 17 69
Drugs Analgesia 18 61
Investigations ABG - 31
Investigations Urea and electrolytes - 44
Investigations Full blood count - 44
A = documented clinical variables pre-intervention; B = documented clinical variables post-intervention; CNS = central nervous system; ABG = arterial blood gas.
Student’s t-test (unpaired) comparing categories A and B: p<0.001.
Table 3. Summary of analysis of the decision nodes
Resuscitation process Decision node
A (compliant, total), N
B (compliant, accurate), n
C (compliant, inaccurate), n
Resuscitation Emergency airway 99 98 1
Resuscitation Oxygen required 85 84 1
Resuscitation Shock present 81 76 5
Resuscitation Active rewarming required 72 68 4
End-point of resuscitation Interpretation of urine output 19 14 5
Drugs Antibiotics administered 56 47 9
Drugs Analgesics administered 61 47 14
Investigations ABG 31 30 1
Investigations Urea and electrolytes 62 53 9
Investigations Full blood count 44 43 1
A = compliance completing decision nodes; B = compliant and accurate; C = compliant and inaccurate; ABG = arterial blood gas.
Student’s t-test (unpaired) comparing categories B and C: p<0.0001.
RESEARCH
438
June 2014, Vol. 104, No. 6 that actively asked whether or not analgesia was indicated. Nodesrelating to the quantification of urine output, the admin istration of antibiotics and analgesics and the interpretation of laboratory results were particularly poorly completed. Fig. 1 shows examples of this cognitive dissonance: the admitting clinicians have recorded low systolic blood pressures and low core body temperature, but have incorrectly selected the ‘shock not present’ and the ‘no need for active rewarming’ tick-boxes, respectively. Compliance with documentation of special investigation results was poor: in 19 cases, abnormal urea and electrolyte results were incorrectly interpreted, and in eight cases abnormal full blood count results were not recognised.
Discussion
Preprinted tick-box forms have been shown to improve comm- unication between units and hospitals, and checklists have been shown to improve safety in the operating room.[7-10] This research has been adopted from the aviation industry, where checklist use is routine and has been successful in promoting safety and reducing error.[9,10] Checklists fulfil a number of functions. They force staff to record specific data, which then fosters interpretation of and reaction to data results. They also promote teamwork and co-operation.
However, checklists need to be implemented within a broader culture of patient safety if they are to be effective. Our experience supports this contention, as while our tick-box admission forms improved documentation, they did not necessarily improve quality of care, our audit revealing persistent violations of safe practice.
Documentation pertaining to the resuscitation process was well recorded, with the exception of the record of core body temperature.
The monitoring of urine output as a guide to resuscitative efforts was poorly captured, as was the need for appropriate drugs. This is a significant failing, as delayed initiation of antibiotics predicts increased morbidity from sepsis. The timeous review of blood results was particularly poorly performed; once again this was a significant omission, as delayed recognition of acute kidney injury translates into increased morbidity. In addition to these limitations in the data capture process, the interpretation of data was problematic.
This misinterpretation of physiological data may be a result of cognitive dissonance, which is the psychological discomfort a person experiences when attempting to reconcile conflicting views of reality simultaneously.[1,2,11] A view of reality is referred to as cognition.
The theory states that people are driven to eliminate a feeling of dissonance by eliminating an existing cognition. In other words, an individual may be biased towards a certain decision, even though the evidence favours an alternative decision. We have previously described the problem of cognitive dissonance in trauma care.[3] This study demonstrates that clinicians can fail to interpret abnormal clinical data. The examples cited in Fig. 1 illustrate the phenomenon of cognitive dissonance.
The major limitation of our tick-box admission forms is that it is a paper-based system. It is possible for clinicians to override (omit) the decision nodes, as there is no mechanical lockout system that forces them to comply. A mechanical lockout system is a generic error- reduction strategy that prevents the next step in a process, unless certain preceding tasks have been completed.[1-3] The most common example of such a system is an online purchase system. It is designed to prohibit completion if certain mandatory data are not entered. The
purchaser is forced to either comply by entering mandatory data or abandon the process completely. We have shown that this pattern is difficult to achieve with a paper-based system, as it cannot overcome the problem of non-compliance and cognitive dissonance. Our research group’s next intended step is to translate the current tick-box admission form into an electronic format. This could theoretically be designed with a mechanical lockout system and function as a clinical decision support system. Such systems include electronic prompts to promote compliance and direction of medical care down appropriate clinical pathways.
Conclusion
Our tick-box admission forms improved the quality of documen- tation, but revealed a significant incidence of violations of safe practice.
Improving clinical care in our environment is a complex endeavour that requires a multifaceted approach with numerous interventions.
A single isolated intervention is unlikely to be successful. Fostering a pervasive culture of patient safety is essential if tick-box admission forms are to be effective in the promotion thereof.
References
1. Reason J. Understanding adverse events: Human factors. Qual Health Care 1995;4(2):80-89. [http://dx.doi.
org/10.1136/qshc.4.2.80]
2. Reason J. Human error: Models and management. BMJ 2000;320(7237):768-770. [http://dx.doi.
org/10.1136/bmj.320.7237.768]
3. Clarke DL, Gouveia J, Thomson SR, Muckart DJ. Applying modern error theory to the problem of missed injuries in trauma. World J Surg 2008;32(6):1176-1182. [http://dx.doi.org/10.1007/s00268-008-9543-7]
4. Laudermilch DJ, Schiff MA, Nathens AB, Rosengart MR. Lack of emergency medical services documentation is associated with poor patient outcomes: A validation of audit filters for pre-hospital care. J Am Coll Surg 2010;210(2):220-227. [http://dx.doi.org/10.1016/j.jamcollsurg.2009.10.008]
5. Alexander T, Fuller G, Hargovan P, Clarke DL, Muckart DJ, Thomson SR. An audit of the quality of care of traumatic brain injury at a busy regional hospital in South Africa. S Afr J Surg 2009;47(4):120-126.
6. Stewart WW, Farina Z, Clarke DL, Thomson SR. Variations in levels of care within a hospital provided to acute trauma patients. S Afr J Surg 2011;49(4):194-198.
7. Goodyear HM, Lloyd BW. Can admission notes be improved by using preprinted assessment sheets? Qual Health Care 1995;4(3):190-193. [http://dx.doi.org/10.1136/qshc.4.3.190]
8. Fisher RB, Dearden CH. Improving the care of patients with major trauma in the accident and emergency department. BMJ 1990;300(6739):1560-1563. [http://dx.doi.org/10.1136/bmj.300.6739.1560]
9. Weiser TG, Haynes AB, Dziekan G, Berry WR, Lipsitz SR, Gawande AA; Safe Surgery Saves Lives Investigators and Study Group. Effect of a 19 item surgical safety checklist during urgent operations in a global patient population. Ann Surg 2010;251(5):976-980. [http://dx.doi.org/10.1097/SLA.0b013e3181d970e3]
10. Haynes AB, Weiser TG, Berry WR, et al.; Safe Surgery Saves Lives Study Group. A surgical safety checklist to reduce morbidity and mortality in a global population. N Engl J Med 2009;360(5):491-499. [http://dx.doi.
org/10.1056/NEJMsa0810119]
11. Clark L. Decision-making during gambling: An integration of cognitive and psychobiological approaches.
Philos Trans R Soc Lond B Biol Sci 2010;365(1538):319-330. [http://dx.doi.org/10.1098/rstb.2009.0147]
Accepted 3 April 2014.
Fig. 1. Examples of cognitive dissonance.
RESEARCH
439
June 2014, Vol. 104, No. 6!
RESEARCH
440
June 2014, Vol. 104, No. 6"
Original research
The introduction of an acute physiological support service for surgical patients is an effective error reduction strategy
D.L. Clarke
a,*, V.Y. Kong
a, L.C. Naidoo
a, H. Furlong
a, C. Aldous
baDepartment of Surgery, Nelson R Mandela School of Medicine, University of Kwa-Zulu Natal, Private Bag 7, Congella, 4013 Durban, South Africa
bResearch Facilitator Post Graduate Committee, Nelson R Mandela School of Medicine, South Africa
a r t i c l e i n f o
Article history:
Received 15 February 2013 Received in revised form 26 April 2013
Accepted 5 June 2013 Available online 21 June 2013 Keywords:
Acute surgical patients High risk
Non cardiac Error prevention
a b s t r a c t
Introduction:Acute surgical patients are particularly vulnerable to human error. The Acute Physiological Support Team (APST) was created with the twin objectives of identifying high-risk acute surgical patients in the general wards and reducing both the incidence of error and impact of error on these patients. A number of error taxonomies were used to understand the causes of human error and a simple risk stratification system was adopted to identify patients who are particularly at risk of error.
Results:During the period November 2012eJanuary 2013 a total of 101 surgical patients were cared for by the APST at Edendale Hospital. The average age was forty years. There were 36 females and 65 males.
There were 66 general surgical patients and 35 trauma patients. Fifty-six patients were referred on the day of their admission. The average length of stay in the APST was four days. Eleven patients were haemo-dynamically unstable on presentation and twelve were clinically septic. The reasons for referral were sepsis,4 respiratory distress,3 acute kidney injury AKI (38), post-operative monitoring (39), pancreatitis,3ICU down-referral,7hypoxia,5low GCS,1coagulopathy.1The mortality rate was 13%. A total of thirty-six patients experienced 56 errors. A total of 143 interventions were initiated by the APST. These included institution or adjustment of intravenous fluids (101), blood transfusion,12 antibiotics,9the management of neutropenic sepsis,1central line insertion,3optimization of oxygen therapy,7correction of electrolyte abnormality,8correction of coagulopathy.2
Conclusion: Our intervention combined current taxonomies of error with a simple risk stratification system and is a variant of the defence in depth strategy of error reduction. We effectively identified and corrected a significant number of human errors in high-risk acute surgical patients. This audit has helped understand the common sources of error in the general surgical wards and will inform on-going error reduction initiatives.
!2013 Surgical Associates Ltd. Published by Elsevier Ltd. All rights reserved.
Article focus
!The peri-operative management of the high risk non cardiac general surgical patient
!Developing a team to manage the acute physiological derangements of acute surgical patients
!Developing an error reduction strategy
Key messages
!Defense in depth strategies may detect error
!There are common patterns of error encountered in the non operative care of acute surgical patients
!Fluid and drug errors are the most common problem
Strengths and limitations of this study’ section.
!This is a single institution study and may not be directly applicable elsewhere
!Pragmatic study that identifies common sources of error in acute care.
*Corresponding author.
E-mail address:[email protected](D.L. Clarke).
Contents lists available atSciVerse ScienceDirect
International Journal of Surgery
j o u r n a l h o m e p a g e : w w w . j o u r n a l - s u r g e r y . n e t
1743-9191/$esee front matter!2013 Surgical Associates Ltd. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijsu.2013.06.003
International Journal of Surgery 11 (2013) 989e992
Publication 14 92
1. Introduction
High-risk operations such as cardiac procedures, solid organ transplantation and major oncological resections are centralized in well resourced centers with highly developed systems of care and are generally performed with acceptable morbidity rates and very low mortality rates.1,2High-risk non-cardiac general and acute care surgery on the other hand is undertaken in a diverse number of hospitals with variable levels of expertise and resources.2e8In the United Kingdom approximately 170,000 patients undergo high- risk, non-cardiac surgery each year and sixty per cent experience significant morbidity and fifteen percent die.2 There is also an increased awareness that human error contributes significantly to these high levels of surgical morbidity and mortality and there is considerable interest in strategies that try and reduce the incidence and limit the impact of human error on health care.2e8 Several modern taxonomies of error have been developed to assist with the understanding of the cause of these errors.3,9,10Any interventions intended to reduce human error associated morbidity and mor- tality amongst acute surgical patients, must accurately quantify surgical risk and ensure that the level of risk is commensurate with the appropriateness of the staff caring for the patient and the re- sources available. To facilitate this the Royal College of Surgeons has defined four levels of care for surgical patients depending on the assessed risk and these are summarized inTable 1.2
Edendale Hospital is a regional hospital in the South African city of Pietermaritzburg in the Province of Kwa-Zulu Natal. It drains patients from the peri-urban settlements around the city and from the four deep rural hospitals of Sisonke district. There are ten Intensive Care Unit (ICU) bedsfive high dependency (HDU) beds in the Pietermaritzburg complex. These units generally run at a hundred percent occupancy and have extremely high patient turnover. For example in 2012 a total of 333 trauma patients were admitted to ICU/HDU in Pietermaritzburg. This is in addition to a significant volume of patients with obstetrical and medical emer- gencies who also require ICU/HDU care. Surgical patients who are not admitted to the ICU or HDU for whatever reason remain under the care of the surgery department. We have run quality improvement programs at Edendale Hospital for the lastfive years.
These programs arose out of a number of audits of the quality of care at our institution, which identified a number of deficits.11,12 These deficits included sub-optimal documentation and poor communication leading to preventable errors and morbidity and even mortality.11,12 In one of these audits a random sample of twenty-five referral letters for patients with Traumatic Brain Injury (TBI) was selected for review. The history was recorded in all the referral letters reviewed, the GCS in 88%, a management plan in 75%, associated localizing signs in 50%, and the condition of the pupils in 13%. In none of the referrals was an assessment of the integrity of the cervical spine recorded. A random sample of 28 inpatient records of patients with TBI was also selected for review.
In 57% of cases the reason for admission was not recorded, in 42% a skull radiograph was omitted despite being indicated, and in 15% a computed tomography (CT) scan was omitted despite the case
meeting our criteria for this investigation. In the management plans of this group there were no recorded orders for supplemental ox- ygen and intravenous (IV) fluids. Clear instructions to perform neurological observations were omitted in all cases. In the obser- vation charts of this group the GCS was recorded in 92%, the state of the pupils was recorded in 71%, pulse rate and blood pressure were documented in 70%, oxygen saturation was only recorded in 42%, and neither blood glucose readings nor core body temperature were ever recorded.11Another audit from our institution revealed that the routine monitoring of acute trauma patients was incon- sistent and incomplete and varied dramatically across geographical locations within the same hospital.12
In response to these audits an Acute Physiological Support Team (APST) was established with the twin objectives of identifying high- risk acute surgical patients who were not admitted to ICU/HDU and reducing the incidence and limiting the impact of human error on these patients. The concept of the APST was based on Reasons
“Swiss Cheese”model of human error.9In this model error is the arrow that has to penetrate multiple layers of defence to strike the patient. However each layer of defence is full of holes, like a piece of Swiss Cheese. If all the holes align then the error can travel in a straight line and hit the patient. Strengthening the multiple layers of defence is known as defence in depth strategy. The APST consisted of a medical officer and an intern under the supervision of a single dedicated senior specialist surgeon. Although there was close liaison with the nursing staff, we were unable to keep a dedicated nursing team and during the period of the study there was considerable turn over of nursing staff. The parent surgical team would formally request that the APST care for the patient. In effect this meant that all patients were seen several times a day by different teams. These included the parent team, the ICU/HDU team and the APST. Each team can be considered as a layer of defence between the error and the patient. The APST hadfifteen male beds and ten female beds.
Each bed was equipped with non-invasive monitoring equipment as well as oxygen points and infusion pumps. We were not able to provide advanced respiratory support such as CPAP, or inotropic support. The APST developed a monthly structured morbidity and mortality meeting as a feedback or closed loop system. Each month the statistics for the team were presented at the morbidity and mortality meeting. The meeting commenced with an overview of the role of the APST and then a brief discussion on modern concepts of error theory. Two index cases of error were selected out and discussed in detail using a published taxonomy of error.
2. Methodology
All patients who were classified as level II or higher according to the Royal College of Surgeons levels of care and were not admitted to ICU or HDU for whatever reason were referred to the APST. A prospective data-base was established to document each patient referred to the APST. Routine demographic data was recor- ded as well as data concerning the clinical course. The APST collated all errors detected by the team and the number of interventions to address these errors initiated by the APST. Error was classified according to type or the process that failed, and cause of error. Types of errors were classified as drug related,fluid related, indwelling device related, management decision related and failure to review in- vestigations. Modern taxonomies of error were used to understand the cause of each error. Errors of omission involved the failure to perform an indicated intervention whereas errors of commission involved the inappropriate application of an inter- vention. Errors of planning involved the incorrect management plan and errors of execution describe an appropriate management plan, which goes awry.
3. Results
During the period October 2012eJanuary 2013 a total of 101 patients were cared for by the APST at Edendale Hospital. The average age was forty years. There were 36 females and 65 males.
There were 66 general surgical patients and 35 trauma patients.
Fifty-six patients were referred on the same day as their admission.
Table 1
Classification of levels of care according to the Royal College of Surgeons of England. (2).
Proposed levels of care (2)
0 Ward Basic observations
1: Enhanced ward At risk of deterioration, more frequent observations, basic resuscitation
2: High dependency Needs detailed observation, intervention or single organ support
3: Intensive care Multiple organ support requiring complex interventions
D.L. Clarke et al. / International Journal of Surgery 11 (2013) 989e992 990